I was reading Steve McConnell's book - Software Estimation. He has discussed about the advantages and disadvantages of Over estimation and under estimation.
Arguments against over estimation - (1) Parkinson's law - work will expand to fill the available time. (2) Goldratt's Student's Syndrome - developers will procrastinate until late in the project and then rush to complete the work.
However I think He missed one important disadvantage of over estimation - Lost opportunity. When we overestimate we are forced to cut off some feature unnecessary or delay the shipping date unnecessary. In either way it is lost opportunity.